Predicting the hardness profile across resistance spot welds in martensitic steels

TLDR: In this article, a simple three-step method was proposed to predict the hardness profile of resistance spot welds using a neural network model, where the temperature history throughout the weld zone is calculated and known models describing the local phase transformations that occur during welding are applied.

Abstract: Much work has been done in the area of modelling resistance spot welding Most of this work predicts the weld geometry resulting from particular process parameters This is well known and can be easily predicted using low-cost tools Predicting post-weld properties is much more difficult though Most of this work needs large computing power and complex modelling techniques such as neural networks This is made more complex when welds exhibit heat-affected zone softening This study proposes a simple three-step method to model the hardness profile of resistance spot welds First, the temperature history throughout the weld zone is calculated Second, known models describing the local phase transformations that occur during welding are applied Finally, the results across the weld are assembled into the hardness profile prediction The validity range of this model is wide in terms of martensite chemical content, sheet thickness and process parameters Predictions were validated against welds in three martensitic steels with varying amounts of carbon and alloying additions